Magnetic coupling and current transducer

Ignoramus7096 wrote: (snip)

I have no problem with your "winding" of the LEM transducer, since it responds to the total current passing through the hole.

But I am a bit worried about your concept of paralleling two current shunts, and expecting the current to divide as you predict. These are very low ohm resistors, and the external connecting resistance may sway the current division, badly.

Thanks.

I am not worried here, if the ratio is not exactly 6:1, it is not the end of the world. The SCR driving board has a trimpot for adjusting current input sensitivity, and my microprocessor can always apply some factor to the voltage readings. As long as the ratio stays constant, I will be in business. The analog ammeter on my welder is using another shunt anyway, for historical reasons.

Are all the connections for the two paralleled shunts brazed of bolted?

Here is a little algebra problem for you: How much change in resistance of the 300 amp shunt (and its end connections) must there be to alter the current through the 50 amp path by 5 amps (10% of full scale) when there is 300 amps total passing through the parallel pair of shunts? I haven't done this, yet, but my gut tells me that it is a very small resistance change. You should be thinking in numbers.

I would think that they will be bolted (with 1/4" and 3/8" bolts).

Suppose that we have two resistors, R1 and R2, and current I.

Then the individual currents will be

I1 = I*R2/(R1+R2) I2 = I*R1/(R1+R2)

Right? If R1 or R2 changes by a small percentage, then I1 and I2 would change by a smaller percentage.

In any case, I should not care too much about their individual resistances and precise ratios. Like I said, I use a wholly different shunt to display analog current reading, anyway (that's how the welder is wired).

With these shunts, I only care that

- the reading on the 50A shunt is a linear function of current

- neither shunt is overloaded.

Output from the current transducer will be fed into two devices:

Both of these are adjustable, the SCR system has a trimpot with adjustable current reading range, and the microprocessor is programmable in BASIC, so I can multiply its reading by a coefficient.

I see your point better, I think. You are saying that over time, bolted connections may change their resistance a little bit, and that could change how my system "sees" current, but by a lot.

Is that what you are saying?

That's why one uses Kelvin connections.

...Jim Thompson

Right. But these two resistors have such low resistance, that I am concerned that the bolted connections will not be small percentages of their resistance, over time and thermal cycles.

A 50 mV 50 A shunt has a resistance of 1 mohm. a 50 mV 300 A shunt has a resistance of 170 uohm.

Are you confident that your bolted connections will be (and remain) a small percentage of 170 uohm?

Yes. When we ordinarily think of connecting a pair of resistors in parallel, we neglect the wiring resistance that connects them. But this may well be a case where the resistance of the connections is higher than that of the resistors, themselves, and varys over time more than the resistance of those resistors. I really don't know what the total resistance of the buss bars and bolt connections between these two shunts would be, but when requirements get down into the uohm region, I start to get very careful.

Would you use the kelvin connections on the 300 ohm shunt to attach the parallel 50 ohm shunt?

If they are heavy enough, I think that might be a good idea, but often these are just small screws for small terminals.

He's talking about the connections for the main current, which is split and goes to two shunts in parallel. Where those conductors connect to the shunts themselves could affect the division ratio.

| | 300A | o----o----o | | ~40A [S1] [S2] ~260A | | o----o----o | etc.

Each of the circles is subject to corrosion, thermal effects, vibration, and all that happy stuff. :-)

Hope This Helps! Rich

This helps me a lot. I can imagine various physical connection schemes that accomplish those circles, with some much worse than others. The best may be to bolt the 300 A line to one side of the 300 amp shunt end, and use the back side of that end to clamp to the branch connection to the 50 amp shunt. It isn't quite a Kelvin connection (across only the calibrated resistance of the 300 amp shunt), but it is close to being one. This eliminates the center circle on the above picture, and forces the actual voltage across the

I see. It would definitely be upsetting for the feedback syustem to "drift" over time due to loosening of connections or corrosion.

(just a week ago, I had to look at why my spa was not heating water, and found a crimped terminal on a 10 ga wire that was loose and had too much resistance. Re-crimping it solved the problem. The lesson is that such things do happen in real life).

I should not care too much, I think, about resistance of conductors and connection points, if it stayed constant. It's the changes that I should worry about.

I will try to do the obvious things, like using split lock washers and tightening screws well and solder on crimps.

I will also be displaying current on the LCD of the controller, as well as showing current on the analog current meter that this CyberTig always had (and which relies on a separate shunt).

I will hopefully be able to see drift if I would compare these two numbers from time to time.

Thanks a lot for your good suggestions. If the above is not sufficient, I will consider redoing it in some way. I have not seen cheap current transducers in the 300A range, yet. I paid about $5 for my current transducer on ebay and I already had a 50A shunt from my junk pile. A 300amp transducer costs $140 at lemusa.com, way too much for me.

Yeppers.

Your drawing is very similar to what I am planning on doing.

(snip)

If you clean the connections on the bars of the shunts and the jumper between 300 and 50 ohm shunts, just before clamping, I think you have a fair chance of success. By the way, I hope we are talking only about DC and low frequency response from the LEM. Otherwise, you also have to worry about the relative inductance of the two paths. I leave it to you to calculate how little inductance there has to be in series with a 170 uohm resistor to double its impedance at, say, 1 kHz. And the LEM will almost certainly raise the series inductance of the 50 A shunt path.

I will definitely use some sandpaper, yes, and I will use split washers also.

Absolutely, yes.

Sure, good point, but I definitely just want to measure DC component.

(I will have a square wave chopper in the same compartment, so interference MAY be an issue, but it would be "downstream" of the shunts.)

Thank you John. My project is considerably closer to the end than to the beginning. My plan is to make this system to "basically work", and then work on polishing and improving my BASIC microcontroller program as I gain experience with this machine.

(snip)

Have you considered belleville washers (domed spring washers)? They provide a large and fairly constant clamping force, in spite of differential thermal expansion between steel clamping bolts and copper clamped bars. I think they provide superior clamping stability than split washers do. Split washers are essentially just thread unscrewing prevention. They do little to help when the joint gets lose because of metal creep (which pure copper tends to do under cyclic stresses). Nylock nuts do as well as split washers without digging chunkks of metal out of the clamped surfaces, and get along very well with belleville washers. ttp://

John, thanks. I have never heard of them before, but I will definitely use them here for constant tension. Thank you!

Wow! _I've_ _helped_ John Popelish!!!!! O frabjous day, calloh callay! Everybody get out your camera and save this one for posterity!

;-D ;-D ;-D

Thanks! Rich